I am a Research Scientist at Google working to enhance human-AI collaboration. My research focuses on developing intelligent systems that can understand and act over long horizons. My primary work includes:

• Multi-Step Interaction: Designing robust AI systems that can interact with humans and the environment over multiple turns to solve complex problems.
• Knowledge Integration: Building and applying knowledge graphs to provide AI with structured memory, enabling better contextual understanding and knowledge injection.
• Multilingual Systems: Ensuring these advanced AI capabilities are accessible and effective across a wide array of languages.

I earned my Ph.D. in Computer Science from the University of Southern California, advised by Dr. Nanyun (Violet) Peng and Dr. Premkumar Natarajan. I was fortunate to have my contributions recognized with an Amazon Alexa Graduate Fellowship, a Best Paper Award at the DLG-AAAI’22 Workshop, and an Area Chair Award at ACL 2023.

Before my doctoral studies, I obtained my B.S. from National Taiwan University under the guidance of Dr. Hung-Yi Lee, where my research focused on speech processing and understanding.

You can download my resume here (Last Update: Aug, 2024).

💥 News and Announcements:

1. [03.2025] Recently published three papers on arXiv: one on LLM tool use, one on memory management for LLM agents, and another on LLM-based data synthesis for event understanding.
2. [08.2024] I joined Google Cloud AI Research as a full-time research scientist.
3. [05.2024] Three papers got accepted at ACL 2024, including two long finding papers and one short finding paper.
4. [05.2024] ⭐ I have successfully defended my Ph.D. ⭐!
5. [03.2024] Our paper Contextual Label Projection for Cross-Lingual Structure Extraction has been accepted to the main conference of NAACL 2024.
6. [09.2023] I joined Google Cloud AI Research as a research intern.
7. [07.2023] One paper got Area Chair Award at ACL 2023.
8. [05.2023] Six papers got accepted at ACL 2023, including four main conference long papers and two long finding papers.